After a PDF statement conversion, misaligned columns show up as amounts sitting in the narration column, debits in the credit column or cells shifted one place to the right. Add a test column to flag every bad row, repair them in bulk, then confirm the row count and totals against the statement.
First, find the damaged rows
Scrolling and squinting does not scale past one page. Put the test in a helper column. Assume Date in A, Narration in B, Debit in C, Credit in D, Balance in E and data from row 2.
| What it catches | Formula in F2 |
|---|---|
| Rows with neither or both of Debit and Credit filled | =IF(COUNT(C2:D2)=1,"ok","check") |
| Amount stored as text, for example 1,250.00 left-aligned | =IF(OR(ISTEXT(C2),ISTEXT(D2),ISTEXT(E2)),"text","ok") |
| Date missing or not a real date | =IF(ISNUMBER(A2),"ok","no date") |
Filter column F for anything that is not ok. A normal transaction has exactly one amount, so COUNT(C2:D2)=1 holds for it. Anything else is a candidate for a slipped cell. Opening and closing balance lines will also be flagged; recognise and ignore them.
Why do the amounts end up in the wrong place?
The converter guesses column boundaries from the gaps between text on the page. When a narration is long, or a debit and a credit are printed close to the column edge, a number can fall on the wrong side of a guessed boundary. The common results are a number glued to the end of the narration, a debit that landed under Credit, or a whole row pushed one cell to the right.
Fixing each pattern
Amount stuck to the end of the narration. If B2 reads RENT TRANSFER 18000.00, pull out the last word with this formula, which finds the position of the final space:
- Amount: =VALUE(SUBSTITUTE(RIGHT(B2,LEN(B2)-FIND("~",SUBSTITUTE(B2," ","~",LEN(B2)-LEN(SUBSTITUTE(B2," ",""))))),",",""))
- Narration without it: =TRIM(LEFT(B2,FIND("~",SUBSTITUTE(B2," ","~",LEN(B2)-LEN(SUBSTITUTE(B2," ",""))))))
Both work by turning the last space into a tilde so that FIND can locate it. They assume the amount is the final word and contains no spaces.
Row shifted right. Select the blank cells that should not be there, right-click, Delete, and choose Shift cells left. Do this for a filtered group of identical damage, not the whole sheet, or you will drag good cells out of place.
Debit and credit swapped. Use the balance as the referee, because it comes from the bank, not from the converter. If E3 is lower than E2 the row was a debit; if higher, a credit. Test it in G3 with =IF(E3<E2,"debit","credit") and compare with whichever column holds the amount. Where they disagree, move the number across.
How do I prove the fix worked?
Finish with a row-count test and a totals test. Say your statement lists 86 transactions and shows total debits of ₹61,350.00 and total credits of ₹78,200.00. In the repaired sheet:
- =COUNT($E$2:$E$500) should return 86, since every transaction has a balance.
- =SUM($C$2:$C$500) should equal 61,350.00.
- =SUM($D$2:$D$500) should equal 78,200.00.
If the count is right but the totals are not, an amount is in the wrong column or was misread; compare running balances row by row. If the count is wrong, rows were lost or duplicated. Either way you now know where to look, rather than finding out in a later month.
A middle path is to repair one clean page and compare it with the others. If page 1 aligns after a single fix and pages 2 to 5 each need a different one, the converter is the weak link, not your sheet. Try a different source or format before spending an evening on row-by-row repairs.
When to stop repairing
If more than a handful of rows per page are damaged, patching costs more time than getting a better source. Download the statement as CSV or Excel if your bank offers it, or convert again with a different tool. SpendixAI is another route: upload the PDF, CSV or Excel statement and review the rows before saving. It cannot promise that every layout reads perfectly, which is why the review step exists.
Useful ideas start with a clearer picture of your own spending.
Explore your reports

